**********
*Open data provided
*variable labels contain data definitions.  See paper for more detail on data collection.
*Note that to protect confidentiality of instructors, class size data is aggregated so that it now represents quartiles.  Paper uses number of students so tables will not replicate exactly.  Qualitiative results are identical.




*Table 2 Correlations

pwcorr specificity constructiveness positivity, sig


*Table 3 Positivity



reg positivity i.studygroup##i.femalefaculty class_size sci socsci,  vce(cluster student_num)
reg positivity i.studygroup##i.femalefaculty class_size c.facultygrades##i.femalefaculty sci socsci,  vce(cluster student_num)
reg positivity i.studygroup class_size femalefaculty facultygrades pellgrant whitestudent female relgrade i.yrs_experience sci socsci,  vce(cluster student_num)
reg positivity i.studygroup##i.femalefaculty c.facultygrades##i.femalefaculty c.class_size##i.femalefaculty sci socsci,  vce(cluster student_num)



*Table 4 Specificity

reg specificity i.studygroup##i.femalefaculty class_size sci socsci,  vce(cluster student_num)
reg specificity i.studygroup##i.femalefaculty class_size c.facultygrades##i.femalefaculty sci socsci,  vce(cluster student_num)
reg specificity i.studygroup class_size femalefaculty facultygrades pellgrant whitestudent female relgrade i.yrs_experience sci socsci,  vce(cluster student_num)
reg specificity i.studygroup##i.femalefaculty c.facultygrades##i.femalefaculty c.class_size##i.femalefaculty sci socsci,  vce(cluster student_num)



